Konbu defines 17 tools with complete JSON schemas and descriptions in Japanese. All tools have proper type definitions and required field declarations. However, descriptions are minimal (10-30 chars), parameter descriptions lack actionable constraints (no ranges, patterns, or format guidance), and output schemas are entirely undocumented. Error handling is absent, no recovery guidance or categorization. Tool composition is sound (single responsibility), but parameter design could be more LLM-friendly (e.g., no enum constraints for status fields, no natural-language identifier support).

• Expand all tool descriptions to 50 - 150 characters. Include WHAT (action), WHEN (use case), and WHAT-IT-RETURNS (key fields). Example: 'Create a new memo with title and optional markdown content. Returns memo_id for reference in updates. Use when user wants to save notes or ideas.'
• Document output schemas for all 17 tools. Specify returned fields, types, and which IDs are needed for chaining (e.g., create_memo returns {id, title, content, tags, created_at}). This enables agents to extract memo_id for update_memo calls.
• Add min/max bounds to numeric parameters: limit (1 - 50), offset (0+). Add format guidance to date/datetime fields with examples: 'due_date: YYYY-MM-DD (e.g., 2025-12-31)'. Add regex or length constraints to string fields.
• Declare status as an enum in update_todo schema: {type: 'string', enum: ['open', 'done']}. This prevents LLM hallucination of invalid statuses.
• Add error handling and recovery guidance. For destructive tools (delete_*), document: 'This action is irreversible. Confirm the ID before calling. If deletion fails, the error will indicate why (e.g., memo not found, permission denied).' Consider adding a dry_run parameter to preview impact.